Serge Masson Pietro Caironi Caterina Fanizza Ralf Thomae Roberto Bernasconi Andrea Noto Roberto Oggioni
Giovanni Stefano Pasetti Marilena Romero Gianni Tognoni Roberto Latini Luciano Gattinoni
Circulating presepsin (soluble CD14 subtype)
as a marker of host response in patients
with severe sepsis or septic shock: data
from the multicenter, randomized ALBIOS trial
Received: 2 September 2014 Accepted: 6 October 2014 Published online: 16 October 2014 Ó Springer-Verlag Berlin Heidelberg and ESICM 2014
On behalf of the ALBIOS Study Investigators.
A complete list of centers and investigators participating in the ALBIOS substudy on biomarkers was published elsewhere (Masson et al.2014, Crit Care 18:R6). S. Masson and P. Caironi contributed equally to the present work.
Take-home message:Presepsin, a soluble fragment of the cluster-of-differentiation marker protein 14 (CD14) involved in pathogen recognition by innate immunity, is an early and accurate predictor of host response and mortality in patients with severe sepsis enrolled in the multicenter ALBIOS trial. Changes in its concentration over time may reflect appropriateness of the antibiotic therapy applied.
Electronic supplementary material The online version of this article
(doi:10.1007/s00134-014-3514-2) contains supplementary material, which is available to authorized users.
S. Masson (
)
) R. Bernasconi R. Latini Department of Cardiovascular Research, IRCCS–Istituto di Ricerche Farmacologiche ‘‘Mario Negri’’, Via Privata Giuseppe La Masa 19, 20156 Milan, Italye-mail: [email protected] Tel.: ?39 0239014615
P. Caironi L. Gattinoni
Dipartimento di Fisiopatologia Medico-Chirurgica e dei Trapianti, Fondazione IRCCS Ca’ Granda-Ospedale Maggiore Policlinico, Universita` degli Studi di Milano, Milan, Italy
P. Caironi L. Gattinoni
Dipartimento di Anestesia, Rianimazione, ed Emergenza Urgenza, Fondazione IRCCS Ca’ Granda–Ospedale Maggiore
Policlinico, Milan, Italy
C. Fanizza M. Romero G. Tognoni Fondazione Mario Negri Sud, Santa Maria Imbaro, Italy
R. Thomae
Mitsubishi Chemical Europe GmbH, Duesseldorf, Germany
A. Noto
Azienda Ospedaliera San Paolo, Milan, Italy
R. Oggioni
Ospedale del Mugello, Borgo San Lorenzo, Italy
G. S. Pasetti
Ospedale Civile S. Giovanni di Dio, Orbetello Scalo, Italy
Abstract Purpose: Presepsin is a soluble fragment of the cluster-of-differentiation marker protein 14 (CD14) involved in pathogen recog-nition by innate immunity. We evaluated the relation between its circulating concentration, host response, appropriateness of
antibiotic therapy, and mortality in patients with severe sepsis. Meth-ods: Plasma presepsin was measured 1, 2, and 7 days after enrollment of 997 patients with severe sepsis or septic shock in the multicenter Albumin Italian Outcome Sepsis (ALBIOS) trial. They were randomized to albumin or crystal-loids. We tested with univariate and adjusted models the association of single measurements of presepsin or changes over time with clinical events, organ dysfunctions, appropri-ateness of antibiotic therapy, and ICU or 90-day mortality. Results: Pre-sepsin concentration at baseline (946 [492–1,887] ng/L) increased with the SOFA score, the number of prevalent organ dysfunctions or failures, and the incidence of new failures of the respiratory, coagulation, liver, and kidney systems. The concentration decreased in ICU over 7 days in patients with negative blood cultures, and in those with positive blood cul-tures and appropriate antibiotic therapy; it increased with inappropri-ate antibiotic therapy (p = 0.0009). Baseline presepsin was independently associated with, and correctly reclas-sified, the risk of ICU and 90-day mortality. Increasing concentrations of presepsin from day 1 to day 2 predicted higher ICU and 90-day mortality (adjusted p \ 0.0001 and 0.01, respectively). Albumin had no
effect on presepsin concentration. Conclusions: Presepsin is an early predictor of host response and mor-tality in septic patients. Changes in
concentrations over time seem to reflect the appropriateness of antibi-otic therapy.
Keywords Severe sepsis Septic shock Prognosis Clinical trial Presepsin
Introduction
Despite recent improvements, the incidence of sepsis in hospitalized patients increases steadily and mortality rates remain unacceptably high [1]. Recent failures with new therapeutic strategies may stem from an incomplete understanding of the underlying biology of sepsis [2]. Considering the variety of clinical phenotypes of severe sepsis and septic shock, we need to understand more accurately the pathophysiology of the disease to propose more personalized therapies. In this context, circulating biomarkers may improve early recognition, optimal selection of appropriate therapies, and early risk stratifi-cation [3–5].
Recognition of pathogen-associated molecular patterns (PAMPs) from infecting microorganisms is mediated, within the innate immune system, by proteins of the Toll-like receptor (TLR) family [6]. The mechanism for the recognition of lipopolysaccharide (LPS), a major PAMP from the outer membrane of most Gram-negative bacteria, involves accessory molecules that facilitate the presenta-tion of LPS to the TLR4–MD2 complexes. CD14 is one of these accessory proteins, expressed on the surface of myelomonocytic cells and as a soluble molecule in the circulation. CD14 is a broad-spectrum affinity component of the innate immunity and detects a variety of ligands associated also with Gram-positive bacteria, fungi, and viruses [7,8]. Acute inflammation may furthermore con-tribute to CD14 up-regulation by tissues other than monocytes [9]. During recognition of PAMPs by soluble CD14, a 13-kDa fragment is formed, named sCD14 subtype (sCD14-ST) or presepsin [10]. Presepsin is emerging as a novel circulating marker of sepsis [10–15]. However, like other candidates in sepsis, this biomarker needs to be val-idated in large, multicenter studies with representative populations [3].
In a pilot study, we previously showed that presepsin has a better prognostic performance than procalcitonin in patients with severe sepsis [16]. Here, we extend our investigations on the clinical role of presepsin assay for monitoring host response and appropriateness of antibiotic therapy to a larger subpopulation of 997 patients enrolled in a multicenter randomized clinical trial that assessed the efficacy of albumin replacement in severe sepsis and septic shock [17]. In particular, we aimed to elucidate (1) the characteristics associated with increased presepsin levels, (2) the clinical value of presepsin monitoring during the
disease, both in relation to the development of complica-tions and to the treatment applied, and (3) prognostic accuracy of presepsin for early and long-term outcomes.
Materials and methods
Study designThe Albumin Italian Outcome Sepsis (ALBIOS) study was a multicenter, open-label trial that enrolled 1,818 patients with severe sepsis admitted to 100 intensive care units (ICUs). The patients were randomly assigned to receive either 20 % albumin and crystalloids or crystal-loids alone, from randomization until day 28 or discharge from the ICU, whichever came first. Study design, inclusion and exclusion criteria, and main results were published elsewhere [17] and are summarized in the Supplementary Appendix.
The study was compliant with the 1975 Declaration of Helsinki as revised in 2008, and approved first by the Institutional Review Board of the Fondazione IRCCS Ca’ Granda-Ospedale Maggiore Policlinico, Milan, Italy (coordinating center), and subsequently by the appropriate institutional review boards of all the other participating centers. Written informed consent was obtained for all participants.
Sample collection and circulating presepsin measurement
In a subset of 997 patients recruited in 40 centers that participated in a predefined biomarkers substudy, venous blood samples were serially collected 1, 2, and 7 days after enrollment (or at ICU discharge, whichever came first), centrifuged, and plasma was shipped on dry ice to a central repository and stored at -70 °C until being assayed. Presepsin was measured with a chemilumines-cent enzyme immunoassay (PATHFAST Presepsin, Mitsubishi Chemical) by personnel blinded to patients’ characteristics. Intra-assay and inter-assay coefficients were 4.4 % at 445 ng/L, and the limit of detection was 20 ng/L. Samples from a subset of 100 patients have already been analyzed with the same method and included in a preliminary publication [16].
Definitions
Severe sepsis was defined as the presence of a proved or suspected infection in at least one site, two or more signs of systemic inflammatory reaction [18], and at least one severe and acute sepsis-related organ dysfunction. Shock at the time of randomization was defined as a score of 3 or 4 for the cardiovascular component of the Sequential Organ Failure Assessment (SOFA) score [19]. New organ failures were defined as a modification of each component of the SOFA score during the study period to 3 or 4 from a baseline of 0, 1, or 2. Acute kidney injury was defined according to the risk, injury, failure, loss, and end-stage kidney injury (RIFLE) criteria [20], based on the daily incremental increases in serum creatinine from baseline. Clinical resolution of the primary infection, if any, was established by the treating physician according to stan-dard clinical signs. Acquired secondary infections were defined as new infections other than the primary infection responsible for the development of severe sepsis, arising during the study, based upon microbiological and clinical assessments, and evaluated by the treating physician.
Appropriateness of antibiotic therapy was based on all the pathogenic microorganisms isolated in either the site or blood cultures. We considered the first-line empirical antibiotic therapy as appropriate when all the microor-ganisms isolated and reported as responsible for the primary infection were susceptible to at least one antibi-otic administered. For this purpose, we considered antibiotic therapy empirically ongoing on day 1, i.e., about 24 h from enrollment. Antibiotic therapy was considered inappropriate if applied after the first day of enrollment, if it did not cover all the pathogens isolated, or when pathogenic microorganisms were resistant. Microorganism susceptibility to antibiotics was defined by a local microbiologist who was unaware of the allo-cation group or treatment applied.
Statistical methods
Categorical variables are presented as proportions, and continuous variables as means (standard deviation) or medians (Q1–Q3), based on the normality of distribution by Kolmogorov–Smirnov test. Differences in clinical characteristics according to tertiles of presepsin concen-tration were analyzed with the Chi-square test or Fisher’s exact test for categorical variables, and with the analysis of variance or non-parametric Kruskal–Wallis test for continuous variables.
We performed multivariable linear regression analysis with backward elimination to identify the clinical deter-minants independently associated with a high log-transformed concentration of presepsin on day 1 (see additional statistical methods in the Supplementary Appendix). The concentrations of presepsin in clinical
subgroups were compared with the Kruskal–Wallis test and trends across tertiles of presepsin concentration were assessed with the Cochran–Armitage test or one-way analysis of variance for categorical and continuous vari-ables, respectively.
Changes in presepsin concentrations over time according to survival status, treatment, and septic shock at baseline were analyzed with two-way analysis of variance for repeated measurements using log-transformed data. Moreover, to account for the hierarchical nature of the data (repeated measurements within patients) and to control simultaneously for the possible confounding effects of the different variables, multivariate multilevel linear regres-sion models were performed [21] including as covariates SOFA score, time, appropriateness of antibiotic therapy, and an interaction term between time and appropriateness of antibiotic therapy. All available measurements at each time point for each patient were considered.
The relation between presepsin concentration and mortality was first assessed with univariate logistic models entering log-transformed concentration of the biomarker as a continuous variable. A backward selection method was applied to select the final model.
Prognostic discrimination of presepsin was evaluated from the area under the receiver-operating characteristic (ROC) curves (AUC), comparing multivariable models with or without presepsin. The increased discriminative value of presepsin for ICU and 90-day mortality was assessed using the category-free net reclassification improvement (cfNRI) and the integrated discrimination improvement (IDI) index [22]. Changes in presepsin concentrations over time were expressed as the difference in log-transformed concentrations. Kaplan–Meier survival curves were shown by tertiles of relative changes and compared with the log-rank test. Both unadjusted and covariate-adjusted logistic regression models were gen-erated to assess potential associations between changes in presepsin concentrations over time and mortality. A two-sided p value of less than 0.05 was deemed statistically significant.
Results
Clinical characteristics related to the presepsin concentration
The concentration of presepsin on day 1 after randomi-zation in the overall cohort of patients was 946 [492–1,887] ng/L (median [Q1–Q3], n = 958). Presepsin was associated with several demographic, clinical, hemodynamic, and laboratory variables (Table 1). Mul-tiple linear regression analysis showed the strongest association of higher presepsin concentrations, by decreasing order of t value, with high serum creatinine
(standardized beta coefficient = 0.34, p \ 0.0001) and bilirubin (0.20, p \ 0.0001), low hemoglobin (-0.13, p\ 0.0001), older age (0.12, p \ 0.0001), reduced urine output (-0.10, p = 0.0005), immunodeficiency (0.08, p = 0.002), and higher serum lactate (0.08, p = 0.007).
The baseline concentration of presepsin was higher in relation to the severity of organ dysfunctions, as denoted by the SOFA score, and the number of prevalent organ dys-functions or failures (Supplementary Fig. 1). Presepsin concentration was significantly higher in patients with bacterial infection (according to the site or blood culture) than in those with negative culture or in those with no
culture available (Supplementary Table 1). Patients with Gram-negative bacterial infections had higher levels of presepsin than those with Gram-positive infections (p = 0.02 for blood culture). Finally, patients with lung infections had lower baseline presepsin levels than patients with abdominal or urinary tract infections (p \ 0.0002). Time course of presepsin and clinical course
during ICU stay
Higher presepsin on day 1 was associated with hemody-namic instability during the first 24 h of treatment:
Table 1 Clinical characteristics according to baseline presepsin concentration
Characteristics All Presepsin
Tertile 1 (\597 ng/L) Tertile 2 (597–1,397 ng/L) Tertile 3 ([1,397 ng/L) P across tertiles n (%) 958 (100.0) 320 (33.4) 320 (33.4) 318 (33.2) Presepsin–median [Q1–Q3] 946 [492–1,887] 388 [293–493] 949 [753–1,124] 2609 [1,901–4,310] – Age, years 66.6 ± 14.7 63.8 ± 15.9 66.8 ± 14.7 69.2 ± 13.0 0.0002 Women, n (%) 395 (41.2) 129 (40.3) 134 (41.9) 132 (41.5) 0.92
Body mass index, kg/m2 26.4 ± 5.6 26.0 ± 5.4 26.8 ± 5.7 26.5 ± 5.6 0.19
Reason for admission to ICU, n (%)
Medical 540 (56.4) 191 (59.7) 166 (51.9) 183 (57.6) 0.12 Elective surgery 67 (7.0) 19 (5.9) 29 (9.1) 19 (6.0) 0.21 Emergency surgery 351 (36.6) 110 (34.4) 125 (39.1) 116 (36.5) 0.47 Preexisting conditions, n (%) Liver disease 14 (1.5) 4 (1.3) 4 (1.3) 6 (1.9) 0.75 COPD 120 (12.5) 42 (13.1) 40 (12.5) 38 (12.0) 0.90
Chronic renal failure 42 (4.4) 3 (0.9) 7 (2.2) 32 (10.1) \0.0001
Immunodeficiency 124 (12.9) 26 (8.1) 49 (15.3) 49 (15.4) 0.007
Congestive or ischemic heart disease 164 (17.1) 45 (14.1) 55 (17.2) 64 (20.1) 0.13
SAPS II score 46 [35–56] 38 [31–48] 46 [36–56] 53 [44–63] \0.0001
Physiological/laboratory variables
Heart rate, beats/min 103.8 ± 20.8 100.9 ± 20.2 105.6 ± 21.0 105.0 ± 20.9 0.0076
Mean arterial pressure, mmHg 74.6 ± 15.0 77.7 ± 14.8 74.9 ± 14.4 71.2 ± 14.9 \0.0001 Mean arterial pressure after 6 h, mmHg 78.3 ± 13.4 81.3 ± 12.5 79.3 ± 13.8 74.3 ± 13.0 \0.0001
Central venous pressure, mmHg 9.9 ± 4.6 9.5 ± 4.4 9.6 ± 4.6 10.8 ± 4.9 0.0003
PaO2/FiO2 2.14 ± 1.14 2.17 ± 1.11 2.11 ± 1.18 2.13 ± 1.12 0.51 Urine output, mL/h 75.8 ± 70.2 88.1 ± 70.8 83.1 ± 74.1 56.1 ± 60.8 \0.0001 Lactate, mmol/L 3.2 ± 2.7 2.5 ± 2.1 3.3 ± 2.9 3.8 ± 3.0 \0.0001 Serum albumin, g/L 24.4 ± 6.2 25.6 ± 6.5 24.0 ± 6.0 23.7 ± 6.1 0.0005 Hemoglobin, g/dL 10.9 ± 2.0 11.4 ± 2.1 10.9 ± 1.9 10.5 ± 1.8 \0.0001 Serum creatinine, mg/dL 1.97 ± 1.62 1.23 ± 0.87 1.82 ± 1.45 2.88 ± 1.91 \0.0001
White blood cells, 103/mm3 13.2 ± 9.5 13.7 ± 8.4 12.1 ± 9.4 13.9 ± 10.6 0.0078
Central venous oxygen saturation, % 72.6 ± 9.8 73.1 ± 8.9 72.7 ± 10.0 72.0 ± 10.4 0.58
Serum bilirubin (mg/dL) 1.38 ± 1.96 0.96 ± 0.87 1.18 ± 1.07 2.00 ± 3.03 \0.0001 Platelet count (109/L) 188.1 ± 130.1 215.7 ± 131.5 181.4 ± 117.9 167.0 ± 135.6 \0.0001 SOFA score 7.0 [5.0–10.0] 5.0 [4.0–7.0] 8.0 [5.0–9.0] 9.0 [7.0–11.0] \0.0001 Shock, n (%) 540 (56.4) 151 (47.2) 191 (59.7) 198 (62.3) 0.0002 Mechanical ventilation, n (%) 760 (79.3) 241 (75.3) 259 (80.9) 260 (81.8) 0.09 Randomized to albumin, n (%) 474 (49.5) 154 (48.1) 154 (48.1) 166 (52.2) 0.49
Positive blood culture, n (%) 271/852 (31.8) 62/276 (22.5) 88/290 (30.3) 121/286 (42.3) \0.0001 Antibiotics at the time of randomization, n (%) 898 (93.7) 297 (92.8) 307 (95.9) 294 (92.5) 0.14 Antibiotics 6 h after randomization, n (%) 954 (99.6) 318 (99.4) 320 (100) 316 (99.4) 0.48 Appropriateness of antibiotic therapy on day 1
according to site culture, n (%)
422/558 (75.6) 119/156 (76.3) 151/199 (75.9) 152/203 (74.9) 0.95 Appropriateness of antibiotic therapy on day 1
according to blood culture, n (%)
211/259 (81.5) 43/58 (74.1) 75/85 (88.2) 93/116 (80.2) 0.09
ICU mortality, n (%) 253 (26.4) 42 (13.1) 74 (33.4) 137 (43.1) \0.0001
patients with higher presepsin had a lower probability of mean arterial pressure of at least 65 mmHg or lactate lower than 2 mmol/L, and a greater probability of being under vasoactive drugs (p \ 0.0001 for all, Supplemen-tary Table 2). Higher presepsin on day 1 was closely associated with a higher incidence of subsequent new organ failures, as denoted by the respiratory, coagulation,
liver, and kidney organ components of the SOFA score (Fig.1).
There were no differences in circulating presepsin concentrations through the first 7 days after enrollment in patients randomized to albumin or to crystalloids (p for time x treatment interaction = 0.40). Consequently, rel-ative changes in presepsin concentration from enrollment were not different in the two experimental arms on day 2 (p = 0.28) or on day 7 (p = 0.53), both in patients with or without shock. Although presepsin concentration was higher in patients with shock than in those without shock at baseline [1,017 (553–2,128) ng/L, n = 540 vs. 793 (426–1,606) ng/L, n = 418, p \ 0.0001] and at sub-sequent time points, the time course was similar (p for time x shock interaction = 0.50).
Appropriateness of antibiotic therapy
Patients with increasing concentrations of presepsin over the first 7 days were less likely to have received early appropriate antibiotic therapy, based on the positivity of either site or blood cultures, than those with decreasing levels over time (Table2). Presepsin concentration ten-ded to decrease on day 7 in patients with negative blood cultures [-17 (-47 to ?44) % vs. day 1, median (Q1– Q3)] and in those with positive blood cultures and appropriate antibiotic therapy [-18 (-53 to ?38) %], whereas it increased in those with positive blood cultures and inappropriate antibiotic therapy [?24 % (-11 to ?199 %), p = 0.002 by Kruskal–Wallis test, Supple-mentary Fig. 2a]. Similar trends were seen for site cultures [-15 (-49 to ?41), -19 (-49 to ?32), and ?5
Fig. 1 Baseline presepsin concentration according to the incidence of new organ failures, defined as a change in each component during the study period to 3 or 4 from a value of 0, 1, or 2 at baseline. Number of patients with new organ-specific failures: respiration (238, 44.7 % of those without organ-specific failure at baseline), coagulation (186, 21.4 %), liver (57, 6.3 %), kidney (157, 21.8 %), cardiovascular (173, 44.0 %). P across categories by Kruskal–Wallis test. Number of patients without prevalent organ failure by organ and by presepsin tertiles (all; tertile 1, tertile 2, tertile 3): respiration (533; 188, 169, 176), coagulation (868; 303, 292, 273), liver (905; 320, 306, 289), kidney (720; 294, 255, 171), cardiovascular (393; 164, 120, 109)
Table 2 Therapeutic procedures and changes in circulating presepsin concentrations over time
Therapeutic variables Relative changes in presepsin concentration (%, day 7–day 1)
Tertile 1 (\-38 %) Tertile 2 (-37 to ?15 %) Tertile 3 ([16 %) P for trend across tertiles Appropriateness of antibiotic therapy based on site cultures
Appropriate therapy on day 1, n (%) 115/145 (79.3) 108/143 (75.5) 99/148 (66.9) 0.02 Total patient-days, n (%) 803/1,542 (52.1) 763/1,514 (50.4) 732/1,583 (46.2) 0.001 Appropriateness of antibiotic therapy based on blood cultures
Appropriate therapy on day 1, n (%) 60/65 (92.3) 47/58 (81.0) 50/70 (71.4) 0.002 Total patient-days, n (%) 397/1,542 (25.7) 332/1,514 (21.9) 356/1,583 (22.5) 0.03 Activated protein C
For at least 2 days, n (%) 4 (1.7) 3 (1.3) 5 (2.1) 0.73
Total patient-days, n (%) 19 (1.2) 17 (1.1) 24 (1.5) 0.48
Tight glycemic control
For at least 3 days, n (%) 177 (74.4) 183 (76.6) 183 (76.6) 0.58
Total patient-days, n (%) 1,136 (73.7) 1,126 (74.4) 1,182 (74.7) 0.52
Corticosteroids
For at least 3 days, n (%) 98 (41.2) 71 (29.7) 100 (41.8) 0.88
Total patient-days, n (%) 562 (36.4) 408 (26.9) 564 (35.6) 0.66
Extracorporeal hemofiltration for sepsis
For at least 2 days, n (%) 8 (3.4) 5 (2.1) 8 (3.4) 0.99
(-40 to ?83) %, respectively, p = 0.02, Supplementary Fig. 2b].
Presepsin and clinical outcomes
Higher presepsin levels on day 1 were associated with most of the tertiary outcomes assessed at post hoc anal-yses in the ALBIOS trial. Patients with higher presepsin on day 1 had a higher incidence of acute kidney injury and renal replacement therapy during the study, a longer ICU stay, longer mechanical ventilation and time to dis-continuation of vasopressor or inotropic agents, as well as a longer duration and a lower degree of resolution of the primary infection (Supplementary Table 3).
Prognostic value of presepsin
The presepsin concentrations remained high over 7 days in patients who died in the ICU but decreased in survivors (p \ 0.0001 for interaction between time and survival, Supplementary Fig. 3).
Higher presepsin levels on day 1 were associated with subsequent mortality (Supplementary Fig. 4). Adjusting for all potential clinical risk factors, presepsin concentration on day 1 was significantly associated with increased mortality in ICU or at 90 days, in all patients, as well in the subgroup of patients with septic shock, whereas in the subgroup without shock it was signifi-cantly associated with increased ICU mortality only (Table 3). Addition of presepsin on day 1 to a clinical model that included all significant risk factors for mortality improved the prognostic accuracy (area under the ROC curve) in particular for patients with shock, as well as the metrics of reclassification (integrated dis-crimination improvement (IDI) or continuous category-free net reclassification improvement (NRI), Supple-mentary Table 4). This correct reclassification was mainly driven by the non-event component of the NRI. There was no significant interaction between presepsin concentration on day 1 and treatment allocation (albu-min or crystalloids) on ICU or 90-day mortality (Supplementary Table 5).
Figure2 shows the Kaplan–Meier survival curves for ICU or 90-day mortality by tertiles of relative changes of presepsin concentration from day 1 to day 2 or from day 1 to day 7. There was a graded increase in the risk of mortality across tertiles of changes (log rank test: p B 0.0003). In adjusted logistic models, increases in presepsin concentration over 2 or 7 days were indepen-dently associated with ICU or 90-day mortality, with the exception of 2-day changes and 90-day mortality in patients with shock (Supplementary Table 6). The same tendency was observed for patients with or without shock.
Discussion
In a preliminary study on a subgroup of 100 patients of the ALBIOS trial, we previously observed that presepsin had a better prognostic value than procalcitonin in patients with severe sepsis or septic shock [16]. Here, in a larger cohort of well-characterized contemporary patients, we found that early presepsin concentration was mainly influenced by advanced age, poor hepatic and renal functions, shock, and impaired tissue perfusion (low mean arterial pressure and hyperlactatemia). Moreover, high presepsin levels follow-ing the first hours of fluid resuscitation were associated with a greater hemodynamic instability, as indicated by a smaller proportion of patients with stable mean arterial pressure (C65 mmHg), or a greater need for vasoactive drugs. These findings may explain the close relationship between pre-sepsin and the number and degree of organ dysfunctions or failures, indicated by the SOFA scores in all its compo-nents. Presepsin concentrations were higher in patients with lower blood platelets, suggesting a possible relationship between this marker and coagulation disorders [23]. Even though presepsin was higher in patients with shock than in those with severe sepsis without shock, we observed only a weak relation with the cardiovascular component of the SOFA score, based on mean arterial pressure and on the doses of vasopressor drugs administered. Presepsin appears therefore to be a good marker of the host response in a variety of its clinical manifestations, encompassing several organ-specific alterations.
The exact mechanisms of membrane CD14 expression [24], presepsin release during sepsis, and its biological function are still largely unknown. CD14 mediates the response to a variety of PAMPs [7] with the simultaneous release of a soluble fragment [25]. We found slightly higher (ca. 25–30 %) levels of presepsin in patients with bacterial infection than in those with negative or unavailable culture. Soluble CD14 also recognizes bac-terial patterns processed by the Toll-like receptor 2 complex (lipoteichoic acid and several lipoproteins), and hence is not restricted to the identification of Gram-neg-ative endotoxins [8]. Accordingly, presepsin may reflect sepsis-induced immune host defense to microorganisms and its consequences on organ function. At the same time, it is worth noting that two of the strongest determinants of higher circulating presepsin concentration were renal and hepatic dysfunctions (high serum creatinine and biliru-bin), suggesting that impaired plasmatic clearance may also contribute to the elevated levels of the marker.
Immunodeficiency was more frequent in patients with elevated presepsin levels. After adjustment for several clinical covariates (age, severity of the overall disease, number of organ dysfunctions) and for the presence of immunosuppression, patients with higher presepsin had a primary infection of longer duration, and a lower inci-dence of resolution, irrespective of the type and site of
infection. These results suggest that higher levels of presepsin, independently of the severity and type of infection per se, may indicate a loss of infection com-partmentalization, or a state of immunoparalysis, leading to a spreading of the related inflammatory reaction, and of the innate immune host response, which may ultimately lead to multiple organ failure and death. Interestingly, CD14? monocytes have been recently considered responsible for the development of immunosuppressive activity through the production of interferon-gamma in human umbilical cord matrix stem cells [26] or through the IL-1b/prostaglandin E2 axis during pregnancy [27]. In our study, we also found that presepsin levels predicted the occurrence of new organ failures, thereby suggesting that it may be a marker of this process, as previously observed for soluble CD14 [28].
Though we did not design the present study to assess a role for presepsin in guiding antibiotic therapy, we eval-uated its time course in relation to the appropriateness of first-line empirical antimicrobial treatment. We found that monitoring presepsin for the first week of treatment may be a good indicator of the adequacy of the antimicrobial therapy. Indeed, patients with decreasing levels of pre-sepsin over 7 days in ICU were more likely to have received an early appropriate first-line empirical antibi-otic therapy on day 1 than those with increasing levels (92 vs. 71 %). Changes in presepsin levels were also associ-ated with the duration of appropriate antibiotic therapy, as estimated by blood or site culture. Interestingly, most of the inappropriate empirical antibiotic therapy was related to infections caused by multidrug-resistant Gram-negative bacteria. These findings suggest that measuring presepsin may help not only in risk stratification after the diagnosis
of sepsis, but also in monitoring the efficacy of the treatment employed, and in guiding its appropriateness.
We also found that presepsin measured on day 1 predicted mortality, independently of age, reasons for ICU admission, SOFA score, hemodynamic and respira-tory variables, laborarespira-tory markers, and appropriateness of antibiotic therapy. The association was stronger in patients with shock. Correct reclassification of the patients was mainly driven by non-events, suggesting that presepsin accurately lowered risk estimates for survivors. Therefore, early assessment of presepsin may help in identifying, among the most severe and fatal manifesta-tions of sepsis, those states that are still reversible, facilitating a more effective second-line and highly intensive clinical treatments. Short-term increases in presepsin concentrations predicted mortality.
In the ALBIOS trial, patients were randomized to either albumin and crystalloids or crystalloids alone for volume replacement [17]. Albumin replacement did not improve survival at 28 and 90 days in the overall popu-lation, but gave small, significant hemodynamic advantages and was more effective in patients with septic shock. In the present study, presepsin concentrations were not significantly different in the two experimental arms, and there was no interaction between median presepsin concentration and randomized treatment on mortality. Thus, in our experimental conditions, we could not find any significant effect of correcting hypoalbuminemia on presepsin levels.
The present study has some limitations. First, it did not include all the patients enrolled in the ALBIOS trial, as only a subgroup of the participating centers were involved in the biomarker substudy. Nonetheless, these centers
Table 3 Presepsin concentration on day 1 and total mortality
Outcome Patients No. events/no. patients (%) OR [95 % CI] p
ICU mortality All 253/958 (26.4) 1.77 [1.48–2.12] \0.0001 Severe sepsis 94/418 (22.5) 1.59 [1.23–2.05] 0.0004 Shock 159/540 (29.4) 2.29 [1.80–2.92] \0.0001 90-day mortality All 369/945 (39.1) 1.77 [1.49–2.11] \0.0001 Severe sepsis 135/409 (33.0) 1.22 [0.92–1.63] 0.17 Shock 234/536 (43.7) 2.17 [1.66–2.69] \0.0001
OR [95 % CI] for an increment of 1 SD unit in log-transformed concentration of presepsin. The covariates were introduced in the multivariable logistic models if they had a p \ 0.20 at univariate analyses. The following covariates were retained in the final models after backward elimination. For ICU mortality, all patients: age, body mass index, chronic obstructive pulmonary disease, immu-nodeficiency, congestive or ischemic heart disease, reason for admission in ICU, SAPS II score, appropriate antibiotic therapy, PaO2/FiO2, platelet count, mechanical ventilation; for ICU
mor-tality, severe sepsis: age, immunodeficiency, heart rate, PaO2/FiO2,
platelet count, mechanical ventilation; for ICU mortality, septic shock: age, chronic obstructive pulmonary disease, body mass index, congestive or ischemic heart disease, PaO2/FiO2, platelet
count, mechanical ventilation. For 90-day mortality, all patients: age, sex, body mass index, chronic obstructive pulmonary disease, immunodeficiency, congestive or ischemic heart disease, reason for admission in ICU, SAPS II score, appropriate antibiotic therapy, PaO2/FiO2, platelet count, mechanical ventilation; for 90-day
mortality, severe sepsis: age, body mass index, mean arterial pressure at baseline or after 6 h, chronic obstructive pulmonary disease, chronic renal failure, immunodeficiency, SAPS II score, SOFA score at baseline, appropriate antibiotic therapy, PaO2/FiO2,
platelet count, urine output; for 90-day mortality, septic shock: age, sex, body mass index, chronic obstructive pulmonary disease, congestive or ischemic heart disease, SAPS II score, appropriate antibiotic therapy, PaO2/FiO2, platelet count
enrolled the majority of the patients randomized in the ALBIOS trial and also in the biomarker substudy, there-fore guaranteeing a representativeness of the disease investigated. Second, this analysis may have been underpowered. Third, as the ALBIOS was a pragmatic trial, the study protocol did not standardize the treatments employed (with the exception of fluid management), though they have been applied by the attending physicians according to the current guidelines for the treatment of severe sepsis and septic shock [29]. Fourth, the clinical value of presepsin has not been compared with other circulating biomarkers. Fifth, presepsin was not assessed on a daily basis over 1 week, precluding a full evaluation of its time course in the acute phase of sepsis.
Conclusions
The present study suggests that presepsin may be a good candidate for accurate and timely prediction of abnormal
host response at an early stage before it becomes too severe and jeopardizes survival. It also gives preliminary indications that presepsin may be suited for monitoring appropriate antibiotic therapy. Although no single circu-lating marker is likely to be a panacea for the clinical management of severe sepsis in view of its complexity and rapid course, these findings are encouraging and should prompt further validation of presepsin in different clinical settings.
Acknowledgments The authors thank Giuseppe Cappelletti and Alessandro Castelli (GEPA Srl, Bollate, Italy) for valuable help with the presepsin assay.
Conflicts of interest SM, PC, RL, and LG received limited research support from Mitsubishi Chemical Europe GmbH, the manufacturer of the presepsin assay. RT is an employee at Mitsu-bishi Chemical Europe GmbH; he had no access to raw data and did not influence the analyses or drafting of the manuscript. The present work was partially presented at EuroMedLab (Milan, May 2013) and at the Annual Meeting of the American Association for Clinical Chemistry (Chicago, July 2014).
Fig. 2 Changes in presepsin concentrations over time and ICU or 90-day mortality. Kaplan–Meier survival curves for ICU or 90-day mortality by tertiles of changes in presepsin concentration from day 1 to day 2 or from day 1 to day 7. The number of patients at risk is
shown below the curves. Tertiles of relative changes: from day 1 to day 2 (1st tertile,\-14.9 %; 2nd tertile, -14.9 to 13.7 %; 3rd tertile, [13.7 %), from day 1 to day 7 (1st tertile, \-38.4 %; 2nd tertile, -38.4 to ?15.8 %; 3rd tertile,[15.8 %). P for log-rank test
Funding The ALBIOS trial was funded by grants from the Italian Medicines Agency (AIFA, grant FARM6JS3R5, 2006) and the Italian Ministry of Health (Ricerca Finalizzata 2011–2012, grant
RF-2011-02348358). Reagents for measuring presepsin were kindly provided by Mitsubishi Chemical Europe GmbH.
References
1. Kaukonen KM, Bailey M, Suzuki S, Pilcher D, Bellomo R (2014) Mortality related to severe sepsis and septic shock among critically ill patients in Australia and New Zealand, 2000–2012. JAMA 311:1308–1316
2. Angus DC, van der Poll T (2013) Severe sepsis and septic shock. N Engl J Med 369:840–851
3. Singer M (2013) Biomarkers in sepsis. Curr Opin Pulm Med 19:305–309 4. Suberviola B, Castellanos-Ortega A,
Ruiz Ruiz A, Lopez-Hoyos M, Santiban˜ez M (2013) Hospital mortality prognostication in sepsis using the new biomarkers suPAR and proADM in a single determination on ICU admission. Intensive Care Med 39:1945–1952 5. Demaret J, Villars-Me´chin A, Lepape
A, Plassais J, Vallin H et al (2014) Elevated plasmatic level of soluble IL-7 receptor is associated with increased mortality in septic shock patients. Intensive Care Med 40:1089–1096 6. Medzhitov R, Janeway CA Jr (2002)
Decoding the patterns of self and nonself by the innate immune system. Science 296:298–300
7. Wright SD (1995) CD14 and innate recognition of bacteria. J Immunol 155:6–8
8. Di Gioia M, Zanoni I (2014) Toll-like receptor co-receptors as master regulators of the immune response. Mol Immunol. pii:S0161–5890(14)00120-5. doi:10.1016/j.molimm.2014.05.008
9. Liu S, Khemlani LS, Shapiro RA, Johnson ML, Liu K et al (1998) Expression of CD14 by hepatocytes: upregulation by cytokines during endotoxemia. Infect Immun 66:5089–5098
10. Shozushima T, Takahashi G, Matsumoto N, Kojika M, Okamura Y et al (2011) Usefulness of presepsin (sCD14-ST) measurements as a marker for the diagnosis and severity of sepsis that satisfied diagnostic criteria of systemic inflammatory response syndrome. J Infect Chemother 17:764–769
11. Liu B, Chen YX, Yin Q, Zhao YZ, Li CS (2013) Diagnostic value and prognostic evaluation of presepsin for sepsis in an emergency department. Crit Care 17:R244
12. Chenevier-Gobeaux C, Trabattoni E, Roelens M, Borderie D, Claessens YE (2014) Presepsin (sCD14-ST) in emergency department: the need for adapted threshold values? Clin Chim Acta 427:34–36
13. Ulla M, Pizzolato E, Lucchiari M, Loiacono M, Soardo F et al (2013) Diagnostic and prognostic value of presepsin in the management of sepsis in the emergency department: a multicenter prospective study. Crit Care 17:R168
14. Vodnik T, Kaljevic G, Tadic T, Majkic-Singh N (2013) Presepsin (sCD14-ST) in preoperative diagnosis of abdominal sepsis. Clin Chem Lab Med
51:2053–2062
15. Endo S, Suzuki Y, Takahashi G, Shozushima T, Ishikura H et al (2012) Usefulness of presepsin in the diagnosis of sepsis in a multicenter prospective study. J Infect Chemother 18:891–897 16. Masson S, Caironi P, Spanuth E,
Thomae R, Panigada M et al (2014) Presepsin (soluble CD14 subtype) and procalcitonin levels for mortality prediction in sepsis: data from the Albumin Italian Outcome Sepsis trial. Crit Care 18:R6
17. Caironi P, Tognoni G, Masson S, Fumagalli R, Pesenti A et al (2014) Albumin replacement in patients with severe sepsis or septic shock. N Engl J Med 370:1412–1421
18. Bone RC, Balk RA, Cerra FB, Dellinger RP, Fein AM et al (1992) Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/ Society of Critical Care Medicine. Chest 101:1644–1655
19. Finfer S, Bellomo R, Boyce N, French J, Myburgh J et al (2004) A comparison of albumin and saline for fluid resuscitation in the intensive care unit. N Engl J Med 350:2247–2256 20. Kellum JA, Bellomo R, Ronco C (2007)
The concept of acute kidney injury and the RIFLE criteria. Contrib Nephrol 156:10–16
21. Singer JD, Willett JB (2003) Applied longitudinal data analysis: modeling change and event occurrence. Oxford University Press, New York
22. Pencina MJ, D’Agostino RB Sr, D’Agostino RB Jr, Vasan RS (2008) Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 27:157–172 23. Ishikura H, Nishida T, Murai A,
Nakamura Y, Irie Y et al (2014) New diagnostic strategy for sepsis-induced disseminated intravascular coagulation: a prospective single-center
observational study. Crit Care 18:R19 24. Glu¨ck T, Silver J, Epstein M, Cao P,
Farber B, Goyert SM (2001) Parameters influencing membrane CD14 expression and soluble CD14 levels in sepsis. Eur J Med Res 6:351–358
25. Yaegashi Y, Shirakawa K, Sato N, Suzuki Y, Kojika M et al (2005) Evaluation of a newly identified soluble CD14 subtype as a marker for sepsis. J Infect Chemother 11:234–238 26. Wang D, Chen K, Du WT, Han ZB,
Ren H et al (2010) CD14? monocytes promote the immunosuppressive effect of human umbilical cord matrix stem cells. Exp Cell Res 316:2414–2423 27. Vacca P, Cantoni C, Vitale M, Prato C,
Canegallo F et al (2010) Crosstalk between decidual NK and CD14? myelomonocytic cells results in induction of tregs and
immunosuppression. Proc Natl Acad Sci USA 107:1123–11918
28. Endo S, Inada K, Kasai T, Takakuwa T, Nakae H, Kikuchi M, Yamashita H, Yoshida M (1994) Soluble CD14 (sCD14) levels in patients with multiple organ failure (MOF). Res Commun Chem Pathol Pharmacol 84:17–25 29. Dellinger RP, Levy MM, Rhodes A,
Annane D, Gerlach H (2013) Surviving Sepsis Campaign: international guidelines for management of severe sepsis and septic shock, 2012. Intensive Care Med 39:165–228
